The human connectome is a description of all connections between neurons in the human brain. Advances in diffusion magnetic resonance imaging (DMRI) make it possible to image macroscopic fibers of axons in living humans, outlining major structural networks in the brain. It is believed that these networks are affected i a wide range of brain disorders and that mapping connectivity, and connectivity deficits, will advance our understanding of normal development and complex brain disorders. While some theoretical limitations of the imaging techniques are known, little work has been done to quantify the actual impact of these limitations on connectivity measurements in the brain. The project will address this problem using in vivo and ex vivo data from the squirrel monkey, acquired at multiple spatial scales. In Aim 1, DMRI estimates of long-range cortical connectivity will be compared to measurements based on neuroanatomical tracer injections. These studies will provide the first quantitative validation of DMRI measurements of cortical connectivity. In Aim 2, sources of error in DMRI connectivity will be determined by comparing DMRI data to axon distributions measured in the same brain and voxel locations using widefield and confocal microscopy. These experiments will determine the most important limitations in practice and evaluate new adaptive strategies for mitigating the leading sources of error. Although DMRI 'connection strength' is increasingly used as an end-point in human studies, its biophysical interpretation is unclear. The project will also identify the biophysical determinants of connectio strength in brain tissue. Aim 3 will focus on DMRI measurements in the cortex. Recent advances in high spatial resolution human imaging have shown that diffusion is significantly anisotropic in the cortex and can be reliably measured. This observation suggests that DMRI could become a useful method for detecting neurodegenerative disease. In Aim 3, DMRI and histological data from the squirrel monkey will be used to test the hypothesis that diffusion properties reflect myeloarchitecture, neuronal density, and total cell density in cortical tissue. Finally, in Aim 4, a web resource will be constructed to make available the project image data, analysis, and visualization tools. Several of the leading approaches to DMRI analysis will be tested in this project, however it is not feasible to include all current and future methods. Instead, we will provide a platform for future advances in connectomics-an atlas of spatially aligned DMRI and microscopy data to allow the neuroimaging community to evaluate and refine novel methods. In summary, the project will remove important barriers to non-invasive assessment of brain connectivity by identifying critical methodological limitations, testing promising solutions, and facilitating validation of advanced algorithms. The results will benefit current and future efforts to understand the connectome in animals and humans.